Qianxun Raises 3B RMB, Spirit Tops Global Benchmarks
💡Embodied AI unicorn raises 3B RMB fast; model beats US rivals on global leaderboards
⚡ 30-Second TL;DR
What Changed
30B RMB funding in 30 days led by Shunwei (Lei Jun) and Yunfeng (Ma Yun)
Why It Matters
Signals investor shift to commercial validation in embodied AI; Qianxun's dual industrial-retail deployments create data flywheels for model scaling.
What To Do Next
Benchmark Spirit v1.5 on RoboChallenge and replicate its zero-shot tasks with your embodied model.
Key Points
- •30B RMB funding in 30 days led by Shunwei (Lei Jun) and Yunfeng (Ma Yun)
- •Spirit v1.5 tops global benchmarks in zero-shot generalization for complex tasks
- •20万+ hours diverse data via 5th-gen wearables, cost 1/10th traditional methods
🧠 Deep Insight
Web-grounded analysis with 10 cited sources.
🔑 Enhanced Key Takeaways
- •千寻智能采用“沿途下蛋”商业策略，通过在京东MALL等真实零售与工业场景部署Moz机器人，实现高频数据采集以反哺模型迭代，形成“数据采集—模型训练—能力提升”的闭环。
- •Spirit v1.5模型摒弃了传统具身智能依赖的“干净”演示数据，转而采用开放式、目标驱动的数据采集范式，通过训练非完美数据，在Scaling Law验证中表现出比传统方法更高的迁移效率。
- •本轮融资是雷军（顺为资本）与马云（云锋基金）在具身智能赛道的首次罕见“同框”领投，标志着该领域从技术探索阶段正式进入由头部产业资本主导的加速分化与洗牌期。
📊 Competitor Analysis▸ Show
| 竞争对手 | 核心模型/产品 | 融资/背景亮点 | 关键差异化 |
|---|---|---|---|
| 千寻智能 | Spirit v1.5 / Moz | 30天融资30亿，雷军+马云领投 | 强调“脏数据”Scaling Law，自研5代可穿戴设备 |
| 银河通用 | 工业/零售机器人 | 2026年3月获25亿融资 | 侧重于通用具身智能机器人平台 |
| 逐际动力 | LimX Dynamics | 2026年2月获2亿美元B轮 | 侧重于运动控制与人形机器人本体 |
| Pi0.5 (美国) | Pi0.5 | 国际标杆模型 | 曾长期霸榜RoboChallenge，被Spirit v1.5超越 |
🛠️ Technical Deep Dive
- •模型架构：采用Vision-Language-Action (VLA) 统一建模框架，将视觉感知、语言理解与动作生成整合在单一决策流程中，减少多模块串联带来的信息损耗。
- •数据引擎：构建多源融合数据引擎，涵盖互联网视频预训练、遥操作数据及自研第5代可穿戴设备采集数据，数据可用性从30%提升至95%。
- •性能指标：Spirit v1.5在RoboChallenge Table30真机评测中，综合得分66.09，任务成功率50.33%，在复杂指令拆解及多任务连续执行维度表现稳定。
- •硬件基础：Moz机器人搭载高功率密度一体化力控关节，配备高精高速WBC（Whole-Body Control）算法，负载自重比达到1:1。
🔮 Future ImplicationsAI analysis grounded in cited sources
⏳ Timeline
📎 Sources (10)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 雷峰网 ↗